Args: X (np.ndarray): Training data of shape[n_samples, n_features] Returns: np.ndarray: shape[n_samples 1),

Args: X (np.ndarray): Training data of shape[n_samples, n_features] Returns: an instance of self

Args: x (np.array): Training data of shape[1, n_features] Currently, only vector of single sample is supported

For Kernel Methods, data is stored in memory and fitting is done at prediction time. Args: X (np.ndarray): Train

Args: X (np.ndarray): Training data of shape[n_samples, n_features] iterations (int): number of EM interations until

Currently, only L2 regularization is implemented. Args: X (np.ndarray): Training data of shape[n_samples, n_features]

Args: logistic_input (np.ndarray): array of shape[n_samples, 1] Returns: np.ndarray: shape[n_samples, 1], lo

Args: x (np.array): Training data of shape[n_features,] Returns: float: predicted value Raises:

Args: X (np.ndarray): Training data of shape[n_samples, n_features] Returns: np.ndarray: shape[n_samples, 1]

Args: X (np.ndarray): Test data of shape[n_samples, n_features] Returns: np.ndarray: shape[n_samples, 1], pr

Attributes: graph (nx.DiGraph): Directed graph which stores tree nodes (int): Current number of nodes in tree

Epanechnikov Kernel

Args: X (np.ndarray): Training data of shape[n_samples, n_features] y (np.ndarray): Target values of shape[n_samples,

Args: X (np.ndarray): Training data of shape[n_samples, n_features] y (np.ndarray): Target values of shape[n_samples,

Gaussian Kernel

Used by add_layer() to get the leaves of the tree.

Partitions the training data at a given node. Traverses the entire down to the indicated node. Args: node_numbe

Note: currenly only works on single vector and not matrices Args: x (np.ndarray): Training data of shape[1, n_features]

Args: x (np.array): Training data of shape[n_features,] Returns: float: predicted value Raises:

Tricube Kernel

Fits Naive Bayes classifier Args: X (np.ndarray): Training data of shape[n_samples, n_features] y (np.ndarra

Fits Naive Bayes classifier Args: X (np.ndarray): Training data of shape[n_samples, n_features] y (np.ndarra

Args: X (np.ndarray): Training data of shape[n_samples, n_features] y (np.ndarray): Target values of shape[n_samples,

Randomly initializes clusers, uses LLyod's algorithm to find optimal clusters. Args: X (np.ndarray): Training da

Randomly initializes clusers, iteratively update cluster center Args: X (np.ndarray): Training data of shape[n_samples,

Randomly initializes clusers, uses LLyod's algorithm to find optimal clusters. Args: X (np.ndarray): Trainin

Args: X (np.ndarray): Training data of shape[n_samples, n_features] y (np.array): Target values of shape[n_samples]

Args: X (np.ndarray): Training data of shape[n_samples, n_features] y (np.ndarray): Target values of shape[n_samples,

Currently, only L2 regularization is implemented. Args: X (np.ndarray): Training data of shape[n_samples, n_features]

Used by parition_data() to get predecessors of a given node (to walk down the tree).

Args: X (np.ndarray): Training data of shape[n_samples, n_features] y (np.ndarray): Target values of shape[n_samples, 1]

Provides a gaussian kernel density at point given a sample. If the KDE of the entire sample space is required, this method can

Args: logistic_input (np.ndarray): array of shape[n_samples, 1] Returns: np.ndarray: shape[n_samples, 1], lo

Args: x (np.array): Training data of shape[n_features,] Returns: float: predicted value Raises:

Args: x (np.array): Training data of shape[n_features,] Returns: float: predicted value Raises:

Args: x (np.array): Training data of shape[n_features,] Returns: float: predicted value Raises:

Args: x (np.array): Training data of shape[1, n_features] Currently, only vector of single sample is supported

Args: x (np.array): Training data of shape[1, n_features] Currently, only vector of single sample is supported

Note: currenly only works on single vector and not matrices Args: x (np.ndarray): Training data of shape[1, n_features]

Note: currenly only works on single vector (one data instance) and not matrices Args: x (np.array): sample d

Args: x1 (np.array): query point of shape[n_features] k (int): number of nearest neighbors to consider Retur

Args: X (np.ndarray): Training data of shape[n_samples, n_features] probability (bool): If True, return probabilities

Computes the distance between x0 and x1 using the DANN metric which is adaptively defined at query locus Args: x1

Used by Fit() to add a single layer at the bottom of the tree

Used by add_layer() to add two children at a leaf in the tree Args: node_number (int): Node in tree which a new split is

weights (np.array): array of weights of shape[n_samples,] to weight each sample in current stump Notes: Meth

residuals (np.array): array of residuals of shape[n_samples,] calculated from the current model Notes: Metho

Args: X (np.ndarray): Training data of shape[n_samples, n_features] y (np.ndarray): Target values of shape[n_samples,

Method to compute average value for all nodes in the tree

Cross Entropy Error

Args: X (np.ndarray): Training data of shape[n_samples, n_features] y (np.array): Target values of shape[n_samples]

Args: X (np.ndarray): Training data of shape[n_samples, n_features] y (np.array): Target values of shape[n_samples]

Args: X (np.ndarray): Training data of shape[n_samples, n_features] y (np.array): Target values of shape[n_samples]

Args: X (np.ndarray): Training data of shape[n_samples, n_features] y (np.ndarray): Target values of shape[n_samples,

Kernel density calssification prediction based on KDE of each class x (np.array): Test data of shape [n_features] gamma: gau

Method to learn split given a data set (inputs) with target values (values)

PRIM algorithm to learn split at node in tree. Maximizes response mean after "boxing off" 90% of data. Args: dat

Local linear regression eliminates bias at boundries of domain. It uses weighted least squares, determining weights from the

Hessian for regulatrized local logistic regression L2 loss Args: theta (np.array): Current lwlr parameters of shape

Average classiication error

Normalizes input data (subtracts mean from each sample)

Mean Square Error

Regression estimate of the target value by weighted averaging of nearby training examples Args: x (np.array): Tr

Partitions the training data at a given node. Traverses the entire down to the indicated node. Args: node_numbe

Partitions inputs based off of a cutoff dictionary which can contain cutoffs for many varialbes (although this feature is currently u

Note: currenly only works on single vector and not matrices Args: x (np.ndarray): Training data of shape[1, n_features]

Note: currenly only works on single vector and not matrices Args: x (np.ndarray): Training data of shape[1, n_features]

Attributes: X (np.ndarray): Training data of shape[n_samples, n_features] X_normalized (np.ndarray): Mean normalized

Attributes: X (np.ndarray): Training data of shape[n_samples, n_features] X_intercept (np.ndarray): Training data of

Attributes: samples (np.ndarray): Data to determine gaussian mixtures for mus (dict): class means {class: [mean1, mea

Attributes: stump_count (int): current number of stumps generated for model stumps (list): list to hold stumps genera

Attributes: tree_count (int): current number of trees generated for model trees (list): list to hold trees generated

Attributes: tree_count (int): current number of trees generated for model trees (list): list to hold trees generated

Attributes: learned (bool): Keeps track of if classifier has been fit class_names (np.ndarray): array of class names.

Attributes: samples (np.ndarray): Data of known target values values (np.ndarray): Known target values for data

Attributes: samples (np.ndarray): Data to be clusters sample_assignments (np.array): cluster assignments for each sam

Attributes: samples (np.ndarray): Data to be clusters sample_assignments (np.array): cluster assignments for each sam

Attributes: X (np.ndarray): Training data of shape[n_samples, n_features] y (np.array): Target values of shape[n_samp

Attributes: X (np.ndarray): Training data of shape[n_samples, n_features] y (np.array): Target values of shape[n_samp

Args: C (float): weight of penalty for misclassified points in soft-margin error if C == 0, then hard-margin SVM

Args: max_iter (int): Maximum number of iterations through PLA before stoping learning_rate (float): PLA step size

Attributes:: _learned (bool): Keeps track of if Linear Regression has been fit _weights (np.ndarray): vector of weigh

Args: alpha (float): degree of regularization, in range(0, 1) Attributes: learned (bool): Keeps track of if

Used by add_layer() to add two children at a leaf in the tree Args: node_number (int): Node in tree which a new split is

Function for selecting a subset of parameters from X, which minimize the in sample error measure. Algorithm acts in a greedy manner, eit

Computes the class average of each node in the tree. Class average is mean of training data that partitions to the node.

Computes the class average of each node in the tree. Class average is the mode of training data that partitions to the node.

Computes the class average of each node in the tree. Class average is the mean of training data that partitions to the node.